Automatic Model Selection

To use the demo, click the next step button, which will trigger an animation to show how the system works.

Explanation of the demo

This demo shows how anomaly detection is performed in our use case. A time series anomaly detector is needed because we want to be able to monitor automatically network traffic data across communities of routers. Some communities see more traffic than other, at different rates also. Different anomaly detection methods might work better than others, so the goal is to automatically find which algorithm (or combination of algorithm rather) will work the best on a specific input.

Demo information

January 2020
Thibaud Perret
Master Thesis Student